The Empirical Analysis Based On Panel Data Of Price Differences Between A And H Share

The price difference between A and H share has been existing for a long time, the departed researches displayed that exchange rate risk did have affection to price differences between A and H share. But few documents make quantitative analysis on the size of its influence to the price differences. the existing empirical analysis based on panel data of price differences between A and H share never use panel data unit root and cointegration test.The estimated model may appear "false return" phenomenon, influence the model estimates effect.By using the GARCH(1,1)-VAR model based on t distribution I measure the exchange rate risk value.With the other factors which may have influence on the price difference between A and H share such as the demand elasticity differences, Liquidity differences, risk/reward differences, Information asymmetry and investor psychological differences and so on, I set an index system to analysis the issue.About The fluidity index among them I adopt a more reasonable effective velocity concept to overcoming the shortages of using sale price differences or pure volume methods to measure the liquidity.Based on a new data interval of January of 2005 to December of 2009, my paper use the 30 listed companies which have been completed crossing-listing before January of 2005 as the sample. Then I calculate the related indexes of daily trading data. But the trading day of different sample companies are different, and useing daily trading data only the denominator will appear to be zero. This paper will use the monthly arithmetic average of the daily trading data for final index variables and join them into panel data model.Using panel unit root test and cointegration test analysis of the six variables indexes, I found four stable variables and two non-stationary variables. Then I made the residuals of Cointegration combination of the two non-stationary variables. In order to avoid Pseudo regression phenomenon, I made Panel data models with all the smooth variables and the residual of Cointegration combination. The F test and Hausman test display using individual fixed effect estimates is appropriate.At the end of my research I found that the indexes of risk/reward differences and Information asymmetry are the same to be Unit roots; Three kinds of co-integration test method which are Kao inspection, Pedroni inspection and Fisher individual joint inspection show that there are cointegration relation between them. And the residual of the cointegration combination is a stable variable; The F test and Hausman test display using individual fixed effect estimates is appropriate; using GARCH (1,1)-VaR model to calculate the value of exchange rate risk is appropriate.By introducing this exchange rate risk proxy variable into panel data model the conclusion is consistent with the theoretical analysis.The exchange rate risk has significantly positive effects to A and H share price difference which validated related research.The hypothesis of demand elasticity differences, Liquidity differences have significant influences on the price difference.The Information asymmetry to a lesser extent has influences on price difference,This is because the investors of A or H share obtain the information different. About companies with larger scale of listed company the investors of H shares acquire relevant company information earlier than mainland investors.The hypothesis of investor psychological differences is not suitable for China's stock market.